Looking for a 24-bit, 4 channel, through hole ADC with I2C interface but cannot find it

Thread Starter

Tyler Bradbury

Joined Mar 18, 2015
13
I have been looking for an ADC with the following parameters:

-24-bit resolution
-4 Channels
-Serial Interface (I2C preferred)
-Through Hole

I cannot find one that fits just these specs anywhere. Does anyone have any experience using a part like this?

Edit: I have been on all the big sites - Digikey, Mouser, LT, TI, Maxim, etc.
 

OBW0549

Joined Mar 2, 2015
3,566
I don't think you're going to satisfy all of those requirements, especially not the through-hole package. Everything I've seen is in TSSOP.
 

WBahn

Joined Mar 31, 2012
26,398
What are you trying to build that needs 24bit resolution? It's a major undertaking to get anywhere near that.
I agree. If you are working with a +/- 2.5V signal range, then 1 lsb corresponds to 300 nV. To put that in perspective, a 1 mA signal running through a 1 mΩ section of PCB track will develop a voltage drop of 1000 nV. Do you REALLY think you can design your circuitry and layout well enough to control all noise sources (both random and systemic) to that kind of level?
 

ebeowulf17

Joined Aug 12, 2014
3,274
I'm assuming the through-hole requirement is because this will be hand-soldered and done on perf-board as opposed to a custom PCB? I find myself constantly frustrated by the limited supply of through hole chips of all sorts compared to their SMD counterparts for the same reasons. What about using one of these to adapt an SMD chip?

https://www.adafruit.com/products/1207

https://www.adafruit.com/products/1208

https://www.adafruit.com/products/1163

https://www.adafruit.com/products/1206

https://www.adafruit.com/products/1162

There are comparable products at Fry's, on Digikey, etc. I just happened to know where to find them quickly at Adafruit.
 

joeyd999

Joined Jun 6, 2011
4,465
What are you trying to build that needs 24bit resolution? It's a major undertaking to get anywhere near that.
I agree. If you are working with a +/- 2.5V signal range, then 1 lsb corresponds to 300 nV. To put that in perspective, a 1 mA signal running through a 1 mΩ section of PCB track will develop a voltage drop of 1000 nV. Do you REALLY think you can design your circuitry and layout well enough to control all noise sources (both random and systemic) to that kind of level?
I use 24 bit Sigma Deltas ADCs often in commercial products. Yes, you have to be careful in design and layout, and you will *never* get 24 solid bits in a single conversion.

*But*, what you do get is a stream of monotonic conversions nicely distributed about a mean (which is very well statistically defined), giving you the opportunity to trade time for ultimate resolution. This is something that is difficult to do with lower resolution SARs.

They are worth the cost, and getting cheaper and more versatile.

Edit: Oh, and @WBahn, Sigma Deltas excel at quashing systemic (synchronous) noise if you understand the noise sources and set up timing appropriately.
 
Last edited:

OBW0549

Joined Mar 2, 2015
3,566
I use 24 bit Sigma Deltas ADCs often in commercial products. Yes, you have to be careful in design and layout, and you will *never* get 24 solid bits in a single conversion.

*But*, what you do get is a stream of monotonic conversions nicely distributed about a mean (which is very well statistically defined), giving you the opportunity to trade time for ultimate resolution. This is something that difficult to do with lower resolution SARs.

They are worth the cost, and getting cheaper and more versatile.
Well put.

People don't use high-resolution sigma-delta ADCs expecting to get extreme DC accuracy; they use them because they need extreme dynamic range. Vibration measurement and analysis, and especially the recording of seismic signals from FBAs, geophones and hydrophones, is one major application area for these chips and one in which DC accuracy is irrelevant, and dynamic range is everything.
 

joeyd999

Joined Jun 6, 2011
4,465
Well put.

People don't use high-resolution sigma-delta ADCs expecting to get extreme DC accuracy; they use them because they need extreme dynamic range. Vibration measurement and analysis, and especially the recording of seismic signals from FBAs, geophones and hydrophones, is one major application area for these chips and one in which DC accuracy is irrelevant, and dynamic range is everything.
Ummm...

S/Ds usually exhibit extreme DC accuracy (within at most a few LSB, and with little or no 1/f noise) and excellent drift capabilities, as well as monotonicity (i.e. no missing codes). Generally, I use them for very low frequency (close to DC) applications. In fact, since an S/D conversion generally takes more time than a SAR, they are slower converters. High frequency data acquisition systems tend to use SARs.
 

joeyd999

Joined Jun 6, 2011
4,465
+1

Dynamic Range
Yes, Mr. Chips. Most of my designs eliminate the need for "tweaking" (via pots or whatever) by using only a portion of the full-scale input range for any given sensor. Variations in sensors are "absorbed" within the dynamic range.
 

joeyd999

Joined Jun 6, 2011
4,465
2^24=16777216 fantastic ! I would like to know which post processing you do with the ADC data.
Usually either a boxcar or FIR filter, depending on the application requirements.

Many S/Ds allow you to set the conversion rate (allowing the time for resolution trade in hardware) eliminating the need for subsequent digital filtering.
 

joeyd999

Joined Jun 6, 2011
4,465
Just FYI, my favorite cost vs. performance converter these days is the TI ADS1242 (or 43 -- more inputs). Though I am always looking for better.
 

nsaspook

Joined Aug 27, 2009
9,703
2^24=16777216 fantastic ! I would like to know which post processing you do with the ADC data.
It depends on the signal source. For a photo-detector signal dominated with Shot_noise a simple averaging filter is effective in improving the S/N ratio of the signal.
http://www.wou.edu/las/physci/poston/ch471/PDF/Lab3key.pdf

We can take a signal that looks like this as raw ADC data http://forum.allaboutcircuits.com/attachments/moon_chart-png.71607/
to this. http://forum.allaboutcircuits.com/attachments/smoone-pdf.73911/
 

WBahn

Joined Mar 31, 2012
26,398
Edit: Oh, and @WBahn, Sigma Deltas excel at quashing systemic (synchronous) noise if you understand the noise sources and set up timing appropriately.
I have no problem believing that. In general you can kill fixed pattern noise with synchronous conversion. Our chips used to have to be run from linear supplies because the switching noise from SMPS was just too hard to kill because we had no way to synchronize to it. Now many of our chips have onboard switchers specifically so that WE control the switching times and can synchronize it to the processing on the chip.
 
Top